Before beginning an A/B test, make sure the page you'll be experimenting with is getting enough traffic. More important than traffic is the conversion rate of the desired outcome of your experiment. With these two pieces of information, you can estimate how many days your experiment will take.
While there are some differing opinions, according to Google's A/B testing tool, Optimize, a test should be live for at least 14 days.
Why 14 Days?
Where did 14 days come from? The reasoning is as follows: the two-week cycle should be completed. As a result, you give some time to weekday and weekend users who exhibit different behaviors in a variety of sectors and treat them equally based on their responses to variations.
While some say a week, two weeks is a reasonable amount of time if there isn't much traffic.
How Can I Calculate My A/B Test
Some tests may take longer than 14 days to complete.
Some believe that traffic is the most important metric influencing test time. While this is not a myth, your conversion rate is the most important metric. The higher your conversion rate, the faster your test will conclude and provide a statistically reliable result. The tests are inedible when both metrics are high.
"It wouldn't be so bad if I knew how long a test would take," she says. If you are one of these people, you can use VWO's tiny calculation tool, which is also an A/B testing tool, on its website.
You can calculate using this tool: https://cxl.com/ab-test-calculator/
- You should enter your weekly traffic > weekly conversions and number of variants (including control)
- In the 2nd part, to the first input on the left, you should enter your baseline conversion rate (control)
How Should You Read This Report?
- The data you enter in the first and second sections will change the table on the right.
- If there is a 22.52 percent increase, the test will be completed in one week. (It is strongly advised that experiments continue to run on a 2-week cycle.)
- If there is a 15.68% effect, the test will be completed in two weeks.
- The lower the effect you want to see in the experiment, the longer it takes to make a decision based on the experiment results.